Optimising plans using genetic programming

C. Henrik Westerberg, John Levine

Research output: Chapter in Book/Report/Conference proceedingConference contribution book

Abstract

Finding the shortest plan for a given planning problem is extremely hard. We present a domain independent approach for plan optimisation based on Genetic Programming. The algorithm is seeded with correct plans created by hand-encoded heuristic policy sets. The plans are very unlikely to be optimal but are created quickly. The suboptimal plans are then evolved using a generational algorithm towards the optimal plan. We present initial results from Blocks World and found that GP method almost always improved sub-optimal plans, often drastically.
LanguageEnglish
Title of host publicationProceedings of the Sixth European Conference on Planning
Place of PublicationPalo Alto
Number of pages3
Publication statusPublished - 21 May 2014
Event6th European Conference on Planning - Toledo, Spain
Duration: 12 Sep 200114 Jul 2016

Conference

Conference6th European Conference on Planning
Abbreviated titleECP 2001
CountrySpain
CityToledo
Period12/09/0114/07/16

Fingerprint

Genetic programming
Planning

Keywords

  • plan optimisation
  • genetic programming
  • heuristic policy sets
  • planning domains
  • heuristics
  • domain independent technique
  • linear plans
  • genetic operators

Cite this

Westerberg, C. H., & Levine, J. (2014). Optimising plans using genetic programming. In Proceedings of the Sixth European Conference on Planning Palo Alto.
Westerberg, C. Henrik ; Levine, John. / Optimising plans using genetic programming. Proceedings of the Sixth European Conference on Planning. Palo Alto, 2014.
@inproceedings{6df2113fb66442dcabaec15d38587b5d,
title = "Optimising plans using genetic programming",
abstract = "Finding the shortest plan for a given planning problem is extremely hard. We present a domain independent approach for plan optimisation based on Genetic Programming. The algorithm is seeded with correct plans created by hand-encoded heuristic policy sets. The plans are very unlikely to be optimal but are created quickly. The suboptimal plans are then evolved using a generational algorithm towards the optimal plan. We present initial results from Blocks World and found that GP method almost always improved sub-optimal plans, often drastically.",
keywords = "plan optimisation, genetic programming, heuristic policy sets, planning domains, heuristics, domain independent technique, linear plans, genetic operators",
author = "Westerberg, {C. Henrik} and John Levine",
year = "2014",
month = "5",
day = "21",
language = "English",
isbn = "9781577356295",
booktitle = "Proceedings of the Sixth European Conference on Planning",

}

Westerberg, CH & Levine, J 2014, Optimising plans using genetic programming. in Proceedings of the Sixth European Conference on Planning. Palo Alto, 6th European Conference on Planning, Toledo, Spain, 12/09/01.

Optimising plans using genetic programming. / Westerberg, C. Henrik; Levine, John.

Proceedings of the Sixth European Conference on Planning. Palo Alto, 2014.

Research output: Chapter in Book/Report/Conference proceedingConference contribution book

TY - GEN

T1 - Optimising plans using genetic programming

AU - Westerberg, C. Henrik

AU - Levine, John

PY - 2014/5/21

Y1 - 2014/5/21

N2 - Finding the shortest plan for a given planning problem is extremely hard. We present a domain independent approach for plan optimisation based on Genetic Programming. The algorithm is seeded with correct plans created by hand-encoded heuristic policy sets. The plans are very unlikely to be optimal but are created quickly. The suboptimal plans are then evolved using a generational algorithm towards the optimal plan. We present initial results from Blocks World and found that GP method almost always improved sub-optimal plans, often drastically.

AB - Finding the shortest plan for a given planning problem is extremely hard. We present a domain independent approach for plan optimisation based on Genetic Programming. The algorithm is seeded with correct plans created by hand-encoded heuristic policy sets. The plans are very unlikely to be optimal but are created quickly. The suboptimal plans are then evolved using a generational algorithm towards the optimal plan. We present initial results from Blocks World and found that GP method almost always improved sub-optimal plans, often drastically.

KW - plan optimisation

KW - genetic programming

KW - heuristic policy sets

KW - planning domains

KW - heuristics

KW - domain independent technique

KW - linear plans

KW - genetic operators

UR - http://www.aaai.org/ocs/index.php/ECP/ECP01/paper/view/7368

M3 - Conference contribution book

SN - 9781577356295

BT - Proceedings of the Sixth European Conference on Planning

CY - Palo Alto

ER -

Westerberg CH, Levine J. Optimising plans using genetic programming. In Proceedings of the Sixth European Conference on Planning. Palo Alto. 2014